• Journal of the Korea Convergence Society
• /
• v.12 no.1
• /
• pp.29-35
• /
• 2021

License plate recognition plays a key role in intelligent transportation systems. Therefore, it is a very important process to efficiently detect the number and character areas. In this paper, we propose a method to effectively detect license plate number area by applying deep learning and semantic image segmentation algorithm. The proposed method is an algorithm that detects number and text areas directly from the license plate without preprocessing such as pixel projection. The license plate image was acquired from a fixed camera installed on the road, and was used in various real situations taking into account both weather and lighting changes. The input images was normalized to reduce the color change, and the deep learning neural networks used in the experiment were Vgg16, Vgg19, ResNet18, and ResNet50. To examine the performance of the proposed method, we experimented with 500 license plate images. 300 sheets were used for learning and 200 sheets were used for testing. As a result of computer simulation, it was the best when using ResNet50, and 95.77% accuracy was obtained.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.6
• /
• pp.819-824
• /
• 2021

As technologies related to image processing such as autonomous driving and CCTV develop, fog removal algorithms using a single image are being studied to improve the problem of image distortion. As a method of predicting fog density, there is a method of estimating the depth of an image by generating a depth map, and various fog features may be used as training data of the depth map. In addition, it is essential to implement a hardware capable of processing high-definition images in real time in order to apply the fog removal algorithm to actual technologies. In this paper, we implement NLCV (Normalize Local Coefficient of Variation), a feature of fog based on coefficient of variation, in hardware. The proposed hardware is an FPGA implementation of Xilinx's xczu7ev-2ffvc1156 as a target device. As a result of synthesis through the Vivado program, it has a maximum operating frequency of 479.616MHz and shows that real-time processing is possible in 4K UHD environment.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.1
• /
• pp.43-50
• /
• 2022

Correlation Power Analysis(CPA) is a sub-channel attack method that measures the detailed power consumption of attack target equipment equipped with cryptographic algorithms and guesses the secret key used in cryptographic algorithms with more than 90% probability. Since CPA performs analysis based on statistics, a large amount of data is necessarily required. Therefore, the CPA must measure power consumption for at least about 15 minutes for each attack. In this paper proposes a method of using a Convolutional Neural Network(CNN) capable of accumulating input data and predicting results to solve the data collection problem of CPA. By collecting and learning the power consumption of the target equipment in advance, entering any power consumption can immediately estimate the secret key, improving the computational speed and 96.7% of the secret key estimation accuracy.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.4
• /
• pp.19-26
• /
• 2022

With the recent development of intelligent transportation systems, car license plate recognition systems are being used in various fields. Such systems need to guarantee real-time performance to recognize the license plate of a driving car. Also, they should keep a high recognition rate even in problematic situations such as small license plates in low-resolution and unclear image due to distortion. In this paper, we propose a real-time car license plate recognition system that improved processing speed using object detection algorithm based on anchor-free method and text recognition algorithm based on Convolutional Neural Network(CNN). In addition, we used Spatial Transformer Network to increase the recognition rate on the low resolution or distorted images. We confirm that the proposed system is faster than previously existing car license plate recognition systems and maintains a high recognition rate in a variety of environment and quality images because the proposed system's recognition rate is 93.769% and the processing speed per image is about 0.006 seconds.

• Journal of the Korea Society for Simulation
• /
• v.19 no.4
• /
• pp.209-218
• /
• 2010

The aim of the study on the tactical object representation techniques in synthetic environment is on acquiring fundamental techniques for detection and tracking of tactical objects, and evaluating the strategic situation in the virtual ground. In order to acquire these techniques, there need the tactical objects' position tracking and evaluation, and an inter-sharing technique between tactical models. In this paper, we study the algorithms on the sensor data fusion and coordinate conversion, proportional navigation guidance(PNG), and pursuit-evasion technique for engineering and higher level models. Additionally, we simulate the position evaluation of tractical objects using the pursuit and evasion maneuvers between a submarine and a torpedo.

• Journal of The Korean Institute of Defense Technology
• /
• v.5 no.2
• /
• pp.17-22
• /
• 2023

In the midst of the development of science and technology based on the 4th industrial revolution, the ROK Army is moving forward with the ARMY TIGER 4.0 system, a ground combat system that combines future advanced science and technology. The system is developing around an AI-based hyper-connected ground combat system, and has mobility, intelligence, and networking as core concepts. Especially, the dronebot combat system is used as a compound word that refers to unmanned combat systems including drones and ground unmanned systems. In future battlefields, it is expected that the use of unmanned and artificial intelligence-based weapon systems will increase. During the transition to a complete unmanned system, it is a very important issue to ensure connectivity individual unmanned systems themselves or between manned and unmanned systems on the battlefield. This paper introduces the Multiplexer Allocation Problem (MAP) for effective command control and communication of UAV/UGV, and proposes a heuristic algorithm. In addition, the performance of the proposed algorithm is analyzed by comparing the solutions and computing time. Also, we discuss future research area for the MAP.

• Journal of the Korea Society of Computer and Information
• /
• v.19 no.7
• /
• pp.71-76
• /
• 2014

There is to be virtually impossible to solve the very large digits of prime number p and q from composite number n=pq using integer factorization in typical public-key cryptosystems, RSA. When the public key e and the composite number n are known but the private key d remains unknown in an asymmetric-key RSA, message decryption is carried out by first obtaining ${\phi}(n)=(p-1)(q-1)=n+1-(p+q)$ and then using a reverse function of $d=e^{-1}(mod{\phi}(n))$. Integer factorization from n to p,q is most widely used to produce ${\phi}(n)$, which has been regarded as mathematically hard. Among various integer factorization methods, the most popularly used is the congruence of squares of $a^2{\equiv}b^2(mod\;n)$, a=(p+q)/2,b=(q-p)/2 which is more commonly used then n/p=q trial division. Despite the availability of a number of congruence of scares methods, however, many of the RSA numbers remain unfactorable. This paper thus proposes an algorithm that directly and immediately obtains ${\phi}(n)$. The proposed algorithm computes $2^k{\beta}_j{\equiv}2^i(mod\;n)$, $0{\leq}i{\leq}{\gamma}-1$, $k=1,2,{\ldots}$ or $2^k{\beta}_j=2{\beta}_j$ for $2^j{\equiv}{\beta}_j(mod\;n)$, $2^{{\gamma}-1}$ < n < $2^{\gamma}$, $j={\gamma}-1,{\gamma},{\gamma}+1$ to obtain the solution. It has been found to be capable of finding an arbitrarily located ${\phi}(n)$ in a range of $n-10{\lfloor}{\sqrt{n}}{\rfloor}$ < ${\phi}(n){\leq}n-2{\lfloor}{\sqrt{n}}{\rfloor}$ much more efficiently than conventional algorithms.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.53 no.4
• /
• pp.37-47
• /
• 2016

The practical constraints on the commercial FPGAs which contain dedicated wide function multiplexers in their slice structure are incorporated with one of the most advanced FPGA mapping algorithms based on the AIG (And-Inverter Graph), one of the best logic representations in academia. As the first step of the mapping process, cuts are enumerated as intermediate structures. And then, the cuts which can be mapped to the multiplexers are recognized. Without any increased complexity, the delay and area of multiplexers as well as LUTs are calculated after checking the requirements for the tree construction such as symmetry and depth limit against dynamically changing mapping of neighboring nodes. Besides, the root positions of multiplexer trees are identified from the RTL code, and annotated to the AIG as AOs (Auxiliary Outputs). A new AIG embedding the multiplexer tree structures which are intentionally synthesized by Shannon expansion at the AOs, is overlapped with the optimized AIG. The lossless synthesis technique which employs FRAIG (Functionally Reduced AIG) is applied to this approach. The proposed approach and techniques are validated by implementing and applying them to two RISC processor examples, which yielded 13~30% area reduction, and up to 32% delay reduction. The research will be extended to take into account the constraints on the dedicated hardware for carry chains.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.5
• /
• pp.414-423
• /
• 2020

In this paper, we propose a Dilated Convolution Gate Linear Unit (DCGLU) to mitigate the lack of sparsity and small receptive field problems caused by the segmentation map extraction process in sound event detection with weak labels. In the advent of deep learning framework, segmentation map extraction approaches have shown improved performance in noisy environments. However, these methods are forced to maintain the size of the feature map to extract the segmentation map as the model would be constructed without a pooling operation. As a result, the performance of these methods is deteriorated with a lack of sparsity and a small receptive field. To mitigate these problems, we utilize GLU to control the flow of information and Dilated Convolutional Neural Networks (DCNNs) to increase the receptive field without additional learning parameters. For the performance evaluation, we employ a URBAN-SED and self-organized bird sound dataset. The relevant experiments show that our proposed DCGLU model outperforms over other baselines. In particular, our method is shown to exhibit robustness against nature sound noises with three Signal to Noise Ratio (SNR) levels (20 dB, 10 dB and 0 dB).

• The Journal of the Korea institute of electronic communication sciences
• /
• v.13 no.6
• /
• pp.1301-1308
• /
• 2018

90/150 CA which are used as key generators of the cipher system have more randomness than LFSRs, but synthesis methods of 90/150 CA are difficult. Therefore, 90/150 CA synthesis methods have been studied by many researchers. In order to synthesize a suitable CA, the analysis of the characteristic polynomial of 90/150 CA should be preceded. In general, the characteristic of polynomial ${\Delta}_n$ of n cell 90/150 CA is obtained by using ${\Delta}_{n-1}$ and ${\Delta}_{n-2}$. Choi et al. analyzed $H_{2^n}(x)$ and $H_{2^n-1}(x)$, where $H_k(x)$ is the characteristic polynomial of k cell 90/150 CA with state transition rule <$10{\cdots}0$>. In this paper, we propose an efficient method to obtain $H_n(x)$ from $H_{n-1}(x)$ and an efficient algorithm to obtain $H_{2^n+i}(x)$ and $H_{2^n-i}(x)$ ($1{\leq}i{\leq}2^{n-1}$) from $H_{2^n}(x)$ by using this method.